Conference Papers 2019

Seminar
Day 2 (10 Oct 2019), Session 5, Agent Based Modelling 1, 11:30 - 13:00

Status
Accepted, documents submitted

Submitted by / Abstract owner
Peter Vovsha

Authors
Peter Vovsha, Gaurav Vyas, Daniel Florian (INRO)

Short abstract
Existing travel models require a substantial revision to accommodate new phenomena such as TNCs, AVs, and MaaS. The paper summarizes authors’ recent experience in several major cities in the US and Canada with the corresponding model revisions.

Abstract
The landscape of travel services today is changing rapidly. Explosive growth of Transportation Network Companies (TNCs) required a substantial revision of mode choice models and handling new forms of intermodal combinations. Autonomous Vehicles (AVs) in near future can have a substantial impact on how private cars are used as well as further revolutionize shared TNC services. Consolidation of different travel services in an integrated system that provides Mobility as a Service (MaaS) may have a substantial impact on private car ownership.
Existing travel models require a substantial revision to accommodate these new phenomena. The paper summarizes authors’ recent experience in several major cities in the US and Canada with the corresponding model revisions. In cases for Phoenix, AZ and Columbus, OH the modeling approach capitalized on the available advanced ABMs. The Vancouver, BC, study utilized a 4-step model where some useful simplified approaches were developed.
The following main model components needed to be revised:
• Car ownership model. Major revision included a general accessibility impacts and interplay between privately owned AVs and shared AVs in TNCs. This required an inclusion of TNC in the accessibility calculation along with car, transit, and non-motorized modes.
• Trip generation & activity pattern model. One effect includes impacts of general accessibility improvement. Another effect includes improved mobility for special population groups (elderly, youth, disable) and related substantial reduction in escorting needs. Additional effect relates to how people view travel and how they could use travel time productively. In terms of in-vehicle time productivity, AVs offer a significant advantage over conventional vehicles (working, reading, texting, etc). Based on the current research, reductions in in-vehicle time coefficient for AVs can range from 15% to 50%.
• Mode choice model. Multiple effects needed to be addressed. Elderly, youth, disable, and other people without a driver license will have access to AVs; auto driver mode availability should not be constrained by age of 16-18. AVs and TNCs become available at any location any time, including non-home locations. This required a complete revision of tour and trip mode choices with a consideration of multi-modal combinations available on the same tour and even for the same trip. TNC and AV can effectively serve as an access/egress mode to transit. Specifically, AVs create a new mode of transportation that combines advantages of Park-and-Ride (PNR) and Kiss-and-Ride (KNR). As the result, and contrary to many previous studies, AVs could shift mode choice in favor of rapid transit. Another new aspect is that a relatively cheap ubiquitous driverless TNCs can substantially increase the (currently negligible) share of this mode. Modeling TNCs required a new approache to forming the utility function for modes with a very different price range.
• Trip distribution & destination choice model. Effects of convenient and productive travel time are not bound to mode choice only but also should be incorporated in the destination choice. If people can use time spent in the car productively, they might be willing to travel more or further. This effect, however, needed to be constrained by the reasonable daily travel time budgets.
• Vehicle routing model. This is a new component that is currently missing in regional travel models but is essential for modeling TNCs and AVs. This model translates person trips into vehicle movements. It also incorporates parking options and tradeoffs with car repositioning that generates Zero Occupancy Vehicle (ZOV) trips. Related “smart city” scenario includes a possibility to remove parking from the city centers and replace it with AV staging areas elsewhere.
• Highway assignment. The expected congestion reduction due to a better use of road capacity is modeled through adjustments of road capacity and speeds that is differentiated by facility type and AV penetration rate.
• Transit assignment. This model underwent a substantial extension of the considered mode combinations due to a deeper integration of TNCs and AVs with mass rapid transit. The modeled scenarios provided a valuable information for transportation and urban planners for reshaping transit stations to facilitate a massive passenger pick-ups and drop-offs. Transit assignment procedures were revised to evaluate response to introduction of TNC and AVs. This incorporates a reasonably detailed strategic-level representation of combined journey travel cost and reveals emergent, network-wide “mode insertions” into transit journeys that provide insight into the interaction of private mobility services and public transit services (for example TNC-Transit-Walk/bike or TNC-Transit-TNC or transit-TNC-transit).
The paper summarizes the experience with these new techniques and the results of the corresponding regional studies.

Programme committee
Transport Models

Topic
Autonomous vehicles – looking beyond the technology.